True, no argument. However, automation done correctly permits higher production volumes with lower error rates at lower marginal cost. Key: Done correctly. Automation exacerbates the price of errors. My optimism is based on Testa's demonstrated success with factory automation.
I recall some pretty horrible results when robotics were deployed by simply automating manual processes. Optimal industrial robotics does require engineering the product(s) specifically to be built that way. Recent processes have had few such mishaps. BMW, for example, has used i production to prototype new materials (especially carbon fiber) with low risk. Ford has deployed first in Europe then in Louisville, KY and other places with the learning curve tiered according to quality of local assembly staff, among other things. VW spent a large fortune building the plant for the Phaeton and Bentley Continental in Dresden, adapting their learning curve to highly complex low volume before high volume less complex vehicles. Audi has designed for robotic processes very extensively with excellent results.
Considering all that Tesla has already done, plus putting Peter Hochholder (ex-Audi) in charge of production, the prospects for good surprises in Model III are high. He knows an enormous amount about high volume differentiated production of price-sensitive vehicles.
Finally, even though "manual processes are highly tolerant to variations" manual processes also have quality inversely correlated to variation. Robotics can handle variations without error, as long as they have been designed and programmed correctly. Tesla is pretty good at that, not quite so good on the manual side.
We shall soon see, won't we? I am excited to see if they can pull this off, nit least because I have two of those III's reserved.